Neuropsychopharmacology (2014) 39, 907–918 & 2014 American College of Neuropsychopharmacology. All rights reserved 0893-133X/14 www.neuropsychopharmacology.org Association of Gamma-Aminobutyric Acid A Receptor a2 Gene (GABRA2) with Alcohol Use Disorder ,1,2,3,4 2 5 6,7 8 1,7,9 Dawei Li* , Arvis Sulovari , Chao Cheng , Hongyu Zhao , Henry R Kranzler and Joel Gelernter 1 2 Department of Psychiatry, School of Medicine, Yale University, New Haven, CT, USA; Department of Microbiology and Molecular Genetics, 3 4 University of Vermont, Burlington, VT, USA; Department of Computer Science, University of Vermont, Burlington, VT, USA; Neuroscience, 5 Behavior, and Health Initiative, University of Vermont, Burlington, VT, USA; Department of Genetics, Geisel School of Medicine, Dartmouth 6 7 College, Hanover, NH, USA; Department of Biostatistics, Yale School of Public Health, New Haven, CT, USA; Department of Genetics, School of 8 Medicine, Yale University, New Haven, CT, USA; Department of Psychiatry, Perelman School of Medicine of the University of Pennsylvania and Philadelphia VAMC, Philadelphia, PA, USA; 9VA Connecticut Healthcare Center, West Haven, CT, USA Gamma-aminobutyric acid (GABA) is a major inhibitory neurotransmitter in mammalian brain. GABA receptor are involved in a number of complex disorders, including substance abuse. No variants of the commonly studied GABA receptor genes that have been associated with substance dependence have been determined to be functional or pathogenic. To reconcile the conflicting associations with substance dependence traits, we performed a meta-analysis of variants in the GABAA receptor genes (GABRB2, GABRA6, GABRA1,andGABRG2 on chromosome 5q and GABRA2 on chromosome 4p12) using genotype data from 4739 cases of alcohol, opioid, or methamphetamine dependence and 4924 controls. Then, we combined the data from candidate gene association studies in the literature with two alcohol dependence (AD) samples, including 1691 cases and 1712 controls from the Study of Addiction: Genetics and Environment (SAGE), and 2644 cases and 494 controls from our own study. Using a Bonferroni-corrected threshold of 0.007, we found strong associations between À 6 À 5 GABRA2 and AD (P ¼ 9 Â 10 and odds ratio (OR) 95% confidence interval (CI) ¼ 1.27 (1.15, 1.4) for rs567926, P ¼ 4 Â 10 and OR ¼ 1.21 (1.1, 1.32) for rs279858), and between GABRG2 and both dependence on alcohol and dependence on heroin (P ¼ 0.0005 and OR ¼ 1.22 (1.09, 1.37) for rs211014). Significant association was also observed between GABRA6 rs3219151 and AD. The GABRA2 rs279858 association was observed in the SAGE data sets with a combined P of 9 Â 10 À 6 (OR ¼ 1.17 (1.09, 1.26)). When all of these data sets, including our samples, were meta-analyzed, associations of both GABRA2 single-nucleotide polymorphisms remained (for rs567926, À 5 À 6 P ¼ 7 Â 10 (OR ¼ 1.18 (1.09, 1.29)) in all the studies, and P ¼ 8 Â 10 (OR ¼ 1.25 (1.13, 1.38)) in subjects of European ancestry and À 6 for rs279858, P ¼ 5 Â 10 (OR ¼ 1.18 (1.1, 1.26)) in subjects of European ancestry. Findings from this extensive meta-analysis of five GABAA receptor genes and substance abuse support their involvement (with the best evidence for GABRA2) in the pathogenesis of AD. Further replications with larger samples are warranted. Neuropsychopharmacology (2014) 39, 907–918; doi:10.1038/npp.2013.291; published online 13 November 2013 Keywords: addiction; gamma-aminobutyric acid receptor; association; meta-analysis; susceptibility gene INTRODUCTION share common genetic risk to some extent (Fu et al, 2002; True et al, 1999; Xian et al, 2008). Variation in the function Alcohol, heroin, opiate, and methamphetamine dependence of the gamma-aminobutyric acid (GABA) receptor subunit (AD, HD, OD, and MD, respectively) are chronic, relapsing genes is hypothesized to have a role in developing the disorders characterized by compulsive seeking, abuse, disorders (Addolorato et al, 2012; Davis and Wu, 2001). tolerance, and physical dependence on the substance in GABA is the major inhibitory neurotransmitter in mamma- question despite adverse social, personal, and legal con- lian brain, where it acts at GABA receptors, which are sequences. The cost of drug abuse has grown to approxi- A ligand-gated multisubunit chloride channels. The GABA mately one trillion dollars per year in the United States A receptors are targets for many clinically important drugs, (Califano, 2007). Family, twin, and adoption studies have such as the benzodiazepines, general anesthetics, and shown that genetic factors have an important role in the barbiturates. At least 16 distinct subunits of GABA etiologies of AD and drug dependence, and these disorders A receptors have been identified. Functional receptors are pentameric, consisting of protein products of several *Correspondence: Dr D Li, Department of Microbiology and subunit classes: alpha, beta, gamma, delta, and rho. The Molecular Genetics, University of Vermont, Burlington, Vermont genes encoding these proteins are organized chromoso- 05405, USA, Tel: 802-656-9838; E-mail: [email protected] mally in clusters, including the gene cluster GABRB2, Received 27 August 2013; revised 9 October 2013; accepted 10 GABRA6, GABRA1, and GABRG2 on chromosome 5q October 2013; accepted article preview online 18 October 2013 and the gene cluster GABRG1, GABRA2, GABRA4, and A genetic meta-analysis DLiet al 908 GABRG1 GABRA2 RAC1P2 COX7B2 GABRA4 GABRB1 Chr. 4p12 rs567926 rs279858 GABRG2 GABRA1 GABRA6 GABRB2 Chr. 5q rs211013 rs211014 rs2279020 rs3219151 rs2229944 Figure 1 Schematic structure of the chromosome 4 and 5 gene clusters. The solid (dark gray) and open rectangles represent the coding and 50 and 30 UTRs, respectively, and their boundaries are the start codon ATG or stop codon TGA. The arrows indicate transcriptional directions of each gene (50–30). Light gray rectangles represent two unrelated genes (chromosome 4p12), whose structures were not shown. The vertical lines above rs numbers indicate the relative positions of the SNPs analyzed in this meta-analysis. To better show the gene structures, the length of each component was adjusted, and thus the rectangle lengths, inter-rectangle distances, and intergenic regions did not reflect their actual proportions. GABRB1 on chromosome 4p12 (McLean et al, 1995; Reich association studies have reported, for example, that the et al, 1998). 5q34 GABAA receptor genes were involved in AD risk in Ethanol acts by binding and altering membrane-bound Finns (Radel et al, 2005), and the 5q33 GABAA receptor ligand-gated and voltage-dependent ion channels (Davies, genes were associated with HD in Chinese subjects (Loh 2003). The GABAA receptor subunits form pentameric et al, 2007). receptor, which is most commonly observed in the postsynaptic cell membrane of the mammalian brain, and Chromosome 4 GABA Receptor Gene Cluster it allows passage for chloride ions in the center of the A pentamer and binding sites for GABA and modulatory Another GABA gene cluster consisting of GABRG1, GABRA2, drugs. Ethanol is believed to affect GABAA receptors and GABRA4,andGABRB1 is located on chromosome 4p12, their subunit composition. Alterations in the GABAA spanning approximately 1302 kbp. The 140-kbp GABRA2 receptor subunits are responsible for GABAergic signaling gene encodes the alpha 2 subunit of the GABAA receptor. changes, which are ultimately associated with chronic Variation in GABRA2 was associated with AD and with brain alcohol use (Weiner and Valenzuela, 2006). Specifically, oscillations (Edenberg et al, 2004). Two synonymous SNPs, the mesolimbic GABAergic neurons act as inhibitory rs279858 in exon 5 and rs567926 in the 30 UTR, have been regulators of dopamine (DA) neurons in the ventral studied most often. tegmental area (VTA) of the brain. However, in the presence The two gene clusters are parts of the numerous subunit of addictive drugs, mesolimbic dopaminergic neurons are isoforms of the GABAA receptors, which differ in agonist activated. For example, cannabinoids in GABAergic neu- affinity, likelihood of opening, conductance, and other rons cause disinhibition (ie, activation) of DA neurons and properties (Cossart et al, 2005). They are also members of eventually lead to increased DA levels in the VTA region the large ‘Cys-loop’ super family of structurally similar and (Luscher and Malenka, 2011). Similar to the mechanism of evolutionarily related ligand-gated ion channels that also action of opioids, alcohol inhibits GABAergic transmission include nicotinic acetylcholine receptors, glycine receptors, in the VTA by promoting GABAA receptor function and and the serotonin-3 (5HT3) receptor. These clusters are hence leads to activation of mesolimbic DA neurons paralogous (Ensembl release 70). Figure 1 shows the (Nestler, 2005). schematic structures of the eight chromosome 5 and 4 cluster genes, including the five genes and seven SNPs meta- Chromosome 5 GABA Receptor Gene Cluster analyzed in this study. Among the eight cluster genes, only A five genes were studied in this meta-analysis because the Four genes, encoding b2, a6, a1, and g2 subunits of the data were insufficient for meta-analysis for the other three GABAA receptor, form a GABA gene cluster on chromo- genes. Supplementary Table 1 summarizes the human some 5. The GABRB2 gene, encoding the b2 subunit and medical diseases and traits that have been reported to be spanning 259 694 base pairs (bp), is located on 5q34. This associated with these genes. receptor mediates the fastest known inhibitory synaptic To reconcile conflicting findings obtained by independent transmission in the central nervous system. The marker research groups in genetic association studies and to rs2229944 is a synonymous single-nucleotide polymorph- elucidate the genetic relationship with substance depen- ism (SNP) in exon 10 of this gene. The GABRA6 gene, dence traits, the current meta-analysis compiled the spanning 16 940 bp, is also located on 5q34. The frequently available genotype data for each of seven SNPs (GABRA1 analyzed SNP rs3219151 is in the 30 untranslated region rs2279020, GABRA2 rs567926 and rs279858, GABRA6 (UTR).